From the Dot-Com Bubble to AGI: Why Businesses That Wait Will Not Survive
The dot-com era showed how quickly technology can turn ambitious businesses into industry leaders. The internet opened global markets, created new business models and changed how companies reached their customers. AI offers the chance to go further: to expand what a business can create, understand and achieve.
The pursuit of artificial general intelligence (AGI) and superintelligence (SI) makes that opportunity extraordinary. At Elvoro, we are building toward SI for marketing, advertising and video production, with safe and ethical use of AI as a guiding principle. The dot-com era helps explain why the companies that put new capabilities to work early can gain advantages that last—and what it takes to become one of them.
Key takeaways
- The dot-com crash ended a market boom, not the internet's commercial transformation—a distinction that matters again as AI advances.
- Superintelligence raise the stakes; the ability to benefit from greater intelligence takes more than access to a model.
- AI high performers organize their work differently. Their practices reveal why technology, oversight and execution belong together.
- Elvoro turns the technology shift into a practical marketing advantage by connecting leading models, proprietary engineering and expert execution.
Key terms, explained
What was the dot-com bubble?
The dot-com bubble was the surge in internet-company valuations in the late 1990s, driven by expectations of extraordinary growth. Many prices rose far beyond what those businesses could justify, and the market crashed in the early 2000s. The internet itself continued to transform commerce—a distinction that matters when assessing the potential of AI.
What is AGI?
AGI stands for artificial general intelligence: AI with the breadth and adaptability to perform a wide range of intellectual tasks at human level or beyond. The ambition is a system that can apply what it learns to unfamiliar problems across fields and act autonomously. Researchers differ on the precise threshold for calling a system AGI.
What's the difference between AI, AGI and SI?
AI, or artificial intelligence, is the umbrella term for systems that perform tasks such as learning, reasoning, analysis and content creation. AGI describes broadly capable intelligence across many fields. SI, or superintelligence, describes intelligence substantially beyond human capabilities across those fields; it is also often referred to as artificial superintelligence (ASI). AGI emphasizes breadth; SI adds a much higher level of capability.
What does Elvoro do?
Elvoro brings Superintelligence (SI) into marketing, advertising and video production. Its enterprise marketing service combines AI systems with marketing experts to create content, manage publishing and use performance analysis to improve future work. Its Strategic AI Agent researches markets, competitors and business information to support better decisions. Together, these offerings connect strategic insight with ongoing execution in a way unlike anything possible before.
Marketing turns a great consumer product into a growing business.
For a consumer business, even an exceptional product needs to be discovered, desired and chosen. Marketing makes that happen: it turns attention into purchases and first-time buyers into loyal customers. When AI makes that work faster, smarter and more economical, it changes who can win the customer.
What determines your advantage
- Content production
New audiences and formats require more production time.
Generate and adapt more content across audiences and formats.
- Speed and cost
Output scales with staff hours, handoffs and overhead.
Automate repeatable work to expand output and reduce production costs.
- Learning and improvement
Manual reporting slows the next round of improvements.
Analyse performance at scale and feed the lessons into the next campaign.
Waiting poses a threat your business
The dot-com crash punished speculation; it did not stop the internet from reshaping commerce. E-commerce kept expanding after the crash. [1][22] The lasting risk was mistaking a fall in share prices for a reason to stand still. Today's AI opportunity is just as practical: more work completed, more knowledge put to use and more value created from the same resources. [11][13]
BCG's 2025 study of 1,250 companies found that its most AI-mature businesses reported roughly five times the revenue benefits and three times the cost savings of laggards. [28] Those gains can create room to reinvest in reaching customers, improving the offer and widening the lead.
Building the capability takes more than adopting a tool: teams need time to change workflows and learn what works. [9][10] A rival that produces more, learns faster and wins customers more efficiently can widen its lead while you catch up. For a consumer business, that can threaten revenue, margins and ultimately survival. The time to build the capability is while it still creates an advantage.
AI leaders are already turning capability into revenue and savings
Realised AI benefits · 2024
- Laggards
- AI leaders
Revenue increase
| Laggards | AI leaders | |
|---|---|---|
| Revenue increase | 1.2% | 6.2% |
Cost reduction
| Laggards | AI leaders | |
|---|---|---|
| Cost reduction | 2.0% | 6.0% |
2024 benefits reported in BCG's 2025 survey of 1,250 companies. AI leaders denotes BCG's most mature, “future-built” cohort. Both measures are percentages of annual revenue in the areas where AI was applied, not necessarily the whole business. The survey shows an association with AI maturity. [28]
[28] BCG, The Widening AI Value Gap (2025), Exhibit 2: realised benefits in 2024.
AGI changes what a business can do. Superintelligence raises the ambition.
The dot-com bubble reflected extraordinary expectations of what the internet would make possible. The enduring business opportunity came from turning connectivity into better ways to sell, serve and operate. AI extends that transformation to intellectual work: research, creative development, reasoning and increasingly the execution of complex tasks. The competitive question is how much more a business can achieve with those capabilities. [8][9][17]
AGI combines human-level capability with breadth across intellectual tasks; superintelligence goes substantially beyond human capability across those fields. Neither label, on its own, tells a business how reliably a system can run a campaign, protect customer information or exercise judgment. Research distinguishes performance, breadth and autonomy for precisely this reason. The opportunity is to build operations that can absorb more powerful intelligence as it arrives. [18]
“I think there has been remarkable progress”
Demis HassabisGoogle DeepMind At the WEF Davos in January 2026, Hassabis maintained roughly a 50% chance of human-level cognitive breadth by the decade's end, while identifying scientific creativity as unfinished work. [2]
“We might have AI that is smarter than any human by the end of this year.”
Elon MuskSpaceXAI Musk's 2026 forecast extended to 2027 at the latest, with intelligence exceeding humanity collectively around 2030–2031. A significant advancement for humanity. [3]
“I would guess that this goes faster than people imagine”
Dario AmodeiAnthropic At the WEF Davos in January 2026, Amodei stood broadly by his 2026–2027 outlook, while allowing that it could take a few years; he sees AI-assisted coding and research accelerating progress. [2]
Their timelines differ, but they raise the same business question: how quickly can an organization turn greater intelligence into better work? Buying access to a new model is fast. Developing customer knowledge, reliable processes and the ability to learn from results takes longer. That is the advantage worth building now. [9][10]
Engineering the next era of marketing intelligence
For a consumer brand, that advantage becomes tangible in marketing: better ideas reach customers sooner, more creative approaches can be explored and each campaign can inform the next. Elvoro engineers that connection: leading AI models and proprietary systems, directed by engineers and marketing experts, create, publish and improve content as a fully managed service.
Expand what your brand can create.
Greater creative range. Faster execution.
AI can help teams develop and refine more ideas with fewer production constraints; creative direction turns that capacity into work worth watching, remembering and buying from.
Make each campaign inform the next.
Customer response becomes practical knowledge.
Connect creative decisions to results: which messages attract relevant customers, which explanations resolve objections and which content helps convert interest into purchases.
Build quality into the operation.
Advanced intelligence. Expert direction.
Integrate research, creation, review, publishing and performance analysis, with clear responsibility for brand standards, factual claims and customer data; safe, ethical execution protects the trust marketing depends on.
The research supports this emphasis on the complete operation. McKinsey's 2025 survey found that AI high performers more often reported capable technology, human validation and AI embedded in business processes. Its 2026 survey reinforces the pattern: nearly three-quarters of high performers reported fundamentally redesigning workflows, compared with one-quarter of other respondents. [25][5]
AI-enabled businesses exceed in almost everything
- AI high performers
- Other respondents
Organizations reporting each practice · 2025
| AI high performers | Other respondents | |
|---|---|---|
| Human-validation processes | 65% | 23% |
| Enabling technology infrastructure | 60% | 23% |
| A clear AI roadmap | 60% | 31% |
| AI embedded in business processes | 58% | 20% |
| Engaged senior leaders | 57% | 33% |
| Iterative AI solution development | 54% | 22% |
Six practices selected from McKinsey's 2025 State of AI survey, Exhibit 14; 109 AI high performers and 1,643 other respondents. High performers reported more than 5% of earnings before interest and taxes attributable to AI and significant value from it. These are associations, not measured effects of individual practices. [25]
Put the frontier of AI to work for your growth
What does “prepare your business for AI” actually mean?
Start with a business outcome and build the operation around it; the discipline is turning better technology into better results, then repeating what works.
Choose the result that matters.
Define the objective—such as profitable customer acquisition—and establish a starting point for cost, quality and performance, so activity does not get mistaken for progress.
Give intelligence the right context.
Bring together product facts, customer questions, brand standards and relevant performance data; decide who can approve claims, publish content and access sensitive information.
Connect execution to learning.
Run the full cycle from research and creation to publishing and analysis, then use the results to improve the next cycle and evaluate new models against real business work.
Frequently asked questions
Don't see the answer you're looking for?
Get in touch.
Should a business wait for AGI or SI before investing in AI?
No. Once we reach full AGI it will be impossible to catch up with your competitors. Start with useful applications now and expand as their value becomes clear. The customer knowledge and working practices you develop can carry forward as models improve; waiting for agreement on an AGI milestone postpones that learning. [9][10]
How can a business tell whether AI is creating real value?
Compare the full cost and quality of the work with a clear starting point. In marketing, track qualified demand, customer acquisition and sales alongside production efficiency; use controlled comparisons where possible to separate AI's contribution from other changes.
Does preparing for AI mean replacing the team?
Not necessarily. AI can increase capacity and automate tasks while people set direction and take responsibility; Elvoro's Marketing for Enterprise brings the technology and expert execution together as a managed service.
How can a brand scale AI without sacrificing quality or trust?
Set clear brand and evidence standards, use approved data, limit access and place expert review where errors would matter most. Elvoro combines proprietary systems with engineers and marketing experts to oversee quality across its enterprise marketing service.
What if we need clearer market insight before expanding marketing?
Start by understanding the customer, competitive alternatives and the decision your business needs to make. Elvoro's Strategic AI Agent monitors relevant sources and company information, then turns them into analysis and actionable strategy.
Can a growing business use advanced AI without building its own AI department?
Yes. A managed service can provide the systems, specialist skills and ongoing execution. Assess what the provider actually delivers, who oversees the work and how performance is reviewed, rather than judging the offer by a model name alone.
What this means for you
Elvoro's Marketing for Enterprise gives your brand the creative power, speed and scale of advanced AI, with the execution handled for you. Proprietary systems and marketing experts create, publish and improve your social media content around your growth objectives. Every campaign is a chance to learn more about your customers. Begin building that advantage now.
Primary sources
27 sources
The report covers agent capabilities and organizational adoption, including the annual AI-use series in Figure 4.3.1.
Additional sources and research (21)
A model with uncertain technology productivity generates bubble-like stock-price paths, with empirical support from U.S. railroad and Internet diffusion periods.
- [8] Timothy F. Bresnahan and Manuel Trajtenberg — General Purpose Technologies: Engines of Growth?Back to citation [8] in the text
The paper characterises general-purpose technologies by broad use, continuing improvement and complementary innovation in downstream sectors.
- [9] Erik Brynjolfsson, Daniel Rock and Chad Syverson — The Productivity J-Curve: How Intangibles Complement General Purpose Technologies (2018)Back to citation [9] in the text
Complementary investments in processes, skills and business models can delay visible productivity gains and distort their measurement while valuable intangible assets are built.
- [10] Erik Brynjolfsson, Lorin M. Hitt and Shinkyu Yang — Intangible Assets: Computers and Organizational Capital (2002)Back to citation [10] in the text
Firms combining computer investment with particular organisational practices showed higher valuations and subsequent productivity than the separate investments alone would suggest.
- [11] Erik Brynjolfsson, Danielle Li and Lindsey Raymond — Generative AI at Work (2025)Back to citation [11] in the text
In the published study of 5,172 customer-support agents, AI assistance increased issues resolved per hour by 15% on average, with gains concentrated among less experienced and lower-skilled workers.
The final 2026 paper reports 12.2% more tasks, 25.1% less completion time and approximately 32% higher human-rated quality on suitable tasks (385 of 758 consultants). Quality gains were 29.9% with GPT-4 alone and 33.9% with a prompt overview. Performance worsened on a separate unsuitable task.
- [13] Shakked Noy and Whitney Zhang — Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence (2023)Back to citation [13] in the text
Among 453 college-educated professionals completing selected writing tasks, ChatGPT reduced average completion time by 40% and raised assessed quality by 18%.
The published study finds rapid adoption alongside much smaller intensive use: 27% of employed respondents used generative AI for work in the previous week in late 2024, while 10% used it every workday.
AI-skilled employment investments are associated with recent productivity growth, with gains concentrated in jobs that build durable firm-specific organisational knowledge.
The paper shows that economy-wide productivity estimates depend on both the fraction of tasks affected and the savings on those tasks, warning against extrapolating from easy-to-measure early applications.
- [17] Tyna Eloundou, Sam Manning, Pamela Mishkin and Daniel Rock — GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models (2023)Back to citation [17] in the text
The paper estimates broad potential task exposure to LLMs and finds that complementary software could expand that exposure beyond standalone model access.
- [18] Meredith Ringel Morris and colleagues — Levels of AGI for Operationalizing Progress on the Path to AGI (2023)Back to citation [18] in the text
The framework treats performance, breadth and autonomy as distinct dimensions, rather than reducing AGI to a single undifferentiated label.
Researchers' aggregate forecasts varied markedly by the capability or occupation-automation question and shifted substantially from the previous survey.
The study uses the human task duration associated with a specified agent success probability to measure progress, making clear that a 50% task horizon is a reliability threshold rather than dependable unsupervised work.
METR judged its newer experiment an unreliable estimate of current AI productivity effects because participation and task selection changed and concurrent agents complicated time measurement.
- [22] U.S. Census Bureau — E-Commerce Retail Sales as a Percent of Total Sales, seasonally adjusted quarterly series (2000–2005 observations), via FRED.Back to citation [22] in the text
The quarterly series records e-commerce's growing share of US retail sales through and after the dot-com crash.
Records the fall in inference prices for models meeting a fixed MMLU benchmark threshold; prices use a 3:1 weighting of input and output tokens.
Historical benchmark comparisons for MMMU, GPQA Diamond and GSM8K; Figures 2.7.2, 2.7.4 and 2.6.2.
- [25] McKinsey — The state of AI in 2025: Agents, innovation, and transformation, Exhibits 1–2.Back to citation [25] in the text
Breakdown of experimentation, piloting and scaling across business functions; complements the separate MIT SMR–BCG survey.
Retrospective task-horizon estimates, uncertainty intervals and release dates; this exhibit covers models released in 2023–2025 only. Downloadable source data accompany the designer assets.
Modeled annual productivity-value ranges by business function; excludes implementation costs.